Draft for Review Only – Not for Distribution

1 5 June 2008 2 Marc P. Hayes 3 Habitat Program, Science Division 4 Washington Department of Fish and Wildlife 5 600 Capitol Way N., Mailstop 43143 6 Olympia, WA 98501-1091 7 360/902-2567; Fax: 360/902-2946 8 [email protected]

9 RH: Hayes et al.• Wood and Terrestrial Salamanders

10 Terrestrial Salamander Wood Utilization in Managed Landscapes: Implications for Forestry

11 Practices

12 MARC P. HAYES1, Washington Department of Fish and Wildlife, Habitat Program, 600 Capitol

13 Way N., Olympia, WA 98501, USA

14 TIMOTHY QUINN, Washington Department of Fish and Wildlife, Habitat Program, 600 Capitol

15 Way N., Olympia, WA 98501, USA

16 TIFFANY L. HICKS, Washington Department of Fish and Wildlife, Habitat Program, 600

17 Capitol Way N., Olympia, WA 98501, USA

18 AIMEE P. MCINTYRE, Washington Department of Fish and Wildlife, Habitat Program, 600

19 Capitol Way N., Olympia, WA 98501, USA

20 MARTIN G. RAPHAEL, Pacific Northwest Research Station, Olympia Forestry Sciences

21 Laboratory, 3625 93rd Avenue S.W., Olympia, WA 98512, USA

22 JAMES G. MACCRACKEN, Longview Timberlands, P.O. Box 667, Longview, WA 98632,

23 USA

24 M. ANTHONY MELCHIORS, Weyerhaeuser Company, 32901 Weyerhaeuser Way S., Federal

25 Way, WA 98063, USA

1 Email: [email protected] Draft for Review Only – Not for Distribution 2 | Hayes et al. 26 ANGELA B. STRINGER, The Campbell Group LLC, One S.W. Columbia Street, Suite 1700,

27 Portland, OR 97258, USA

28 ABSTRACT We studied the 4 species of terrestrial salamander in managed forests of

29 southwestern Washington to better understand their use of dead wood as habitat. During April-

30 June, we intensively sampled 10 2-m wide belts oriented perpendicular to stream axes at 14

31 streams in 2001 and 5 streams in 2003. We partitioned belts into macrohabitats (banks and

32 uplands), and characterized habitat within each belt and microhabitats around each salamander.

33 All 4 species were associated with wood, but at different levels, a pattern that seemed related to

34 both spatial (macrohabitat) and temporal (year) patterns in temperature and moisture. Dunn’s and

35 Van Dyke’s salamanders (Plethodon dunni and P. vandykei), the 2 more hydrophilic species,

36 occurred most frequently in the wetter macrohabitat (banks) regardless of year, but also used

37 locally wetter and (in the case of Dunn's salamander) cooler microhabitats during the drier

38 (2001) but not the wetter (2003) year. Though Dunn’s salamander used wood as cover less

39 frequently than Van Dyke’s salamander, its use of wood increased 2.5 times in the drier versus

40 wetter year. Van Dyke’s salamander stood out from the other species in its greater use of large

41 wood, though data were only available for the dry year. In contrast, the 2 less hydrophilic

42 species, ensatina (Ensatina eschscholtzii) and western red-backed salamander (P. vehiculum)

43 were widespread across macrohabitats; and though they exhibited no consistent change in their

44 use of wood under drier conditions, they both selected locally cooler and (in the case of the

45 western red-backed salamander) wetter microhabitats in the drier year. Wood use by terrestrial

46 amphibians is clearly species-specific and an association between the more hydrophilic species

47 and wood, that appears most apparent in dry years, merits further investigation. If Van Dyke’s Draft for Review Only – Not for Distribution 3 | Hayes et al. 48 salamander is dependent on large wood, it may be susceptible to practices that reduce the quality

49 (size) and quantity of that wood, particularly on streambanks.

50 KEY WORDS coarse woody debris, Ensatina eschscholtzii, habitat use, managed landscapes,

51 Plethodon dunni, Plethodon vandykei, Plethodon vehiculum, terrestrial salamanders, timber

52 harvest, Washington.

53 The Journal of Wildlife Management: 00(0): 000-000, 200X

54 Wood debris is believed to be an important resource for many forest-dwelling organisms

55 (Bunnell et al. 1999, Ódor et al. 2001, Bull 2002, Bunnell et al. 2002, Grove 2002, Mathieu et al.

56 2005), including amphibians (Aubry et al. 1988, Aubry and Hall 1991, Corn and Bury 1991,

57 Gilbert and Allwine 1991, Walnick 1997, Butts and McComb 2000). This belief has been based

58 on the frequency with which amphibians were associated with logs (e.g., Corn and Bury 1991),

59 correlations of amphibian abundance with the volume of coarse wood debris (e.g., Butts and

60 McComb 2000), or more limited data on nest site occurrence in downed wood (Hanlin et al.

61 1978, Norman and Norman 1980, Norman 1986, Jones 1989, Blessing et al. 1999, Nauman et al.

62 1999, Olson et al. 2006). Yet, recent work in managed forest suggested that abundance of

63 amphibians was unrelated to the amount of coarse wood debris (Aubry 2000) and adds to the

64 increasing number of studies that demonstrate conflicting results about the importance of wood

65 to terrestrial salamanders (compare Dupuis et al. 1995, Dupuis 1997, Dupuis and Bunnell 1999,

66 and Grialou et al. 2000 to Corn and Bury 1991 and Aubry 2000).

67 We envision 3 major biological reasons why a particular study would fail to show that

68 wood is important to terrestrial amphibians: 1) wood is not important habitat to amphibians and

69 hence, has little effect on suitability of their habitat; 2) wood becomes increasingly important as

70 other types of physical habitat become more limited; and 3) wood becomes important only under Draft for Review Only – Not for Distribution 4 | Hayes et al. 71 certain climatic conditions. The first reason, while plausible, fails to explain why studies of the

72 same species have reached completely opposite conclusions about the wood-amphibian

73 relationship. The second reason has been the subject of speculation. For example, some have

74 suggested that wood and alternative substrates may substitute for one another (Hagar et al. 1995,

75 Bunnell et al. 1997), but the relative value of different substrates to amphibians, at least in the

76 Pacific Northwest, has garnered little attention. The third reason is based in part on distinctive

77 properties of wood and the life history of terrestrial amphibians in the Pacific Northwest. Wood

78 has a substantial capacity to absorb and retain water (Stamm 1935, Stamm and Loughborough

79 1942, Jayme 1958), and thus may create microhabitats that differ from many other substrates,

80 particularly non-porous rocks. Terrestrial amphibians in the Pacific Northwest are exclusively

81 lungless (plethodontid) salamanders (Jones et al. 2005) and rely almost exclusively on moist

82 skin-based gas exchange (Feder and Burggren 1992). Wood may provide hydric or thermal

83 advantages over alternative substrates (Heatwole 1962, Bunnell et al. 2002). In the Pacific

84 Northwest, where salamanders typically restrict their activities to habitats and seasons where risk

85 of desiccation is minimized (Aubry et al. 1988, Grialou et al. 2000, Aubry 2000), availability of

86 wood may be especially important during the summer dry periods.

87 Our overarching purpose was to provide basic data to clarify the relationship between

88 terrestrial amphibians and wood in managed forest landscapes. Our study was structured around

89 Dunn’s salamander (Plethodon dunni) but we included data on Van Dyke’s salamander (P.

90 vandykei), ensatina (Ensatina eschscholtzii) and western red-backed salamander (P. vehiculum).

91 To address this need, we first describe the habitat relationships among the terrestrial salamanders

92 that occur in southwestern Washington State, focusing on the distribution of animals relative to

93 streams, and then on more fine-scale moisture, temperature and wood use patterns. We identify Draft for Review Only – Not for Distribution 5 | Hayes et al. 94 how terrestrial salamander use of wood varied between habitats and years with different moisture

95 and temperature conditions, and suggest how individual salamander species might be influenced

96 by losses of wood associated with forest management practices.

97

98 STUDY AREA

99 The study area was located in the Willapa Hills of southwestern Washington (Figure 1),

100 which comprise the northern segment of the Coast Ranges physiographic province (Franklin and

101 Dyrness 1988). This region had a complex topography of mostly low hills (maximum elevation

102 948 m); valleys with significant alluvial area occurred in only a few larger streams (e.g.,

103 Chehalis, Naselle, and Willapa Rivers; Figure 1). Geology, a complex mix of Tertiary

104 formations, included mostly marine sedimentary (mudstone, siltstones, sandstones) and intrusive

105 basalt formations (Franklin and Dyrness 1988). Over 95 percent of the region was managed for

106 timber. We chose this area because it is the only part of Washington State in which Dunn’s

107 salamander was known to occur (Dvornich et al. 1997).

108

109 METHODS

110 Site Selection

111 We selected sites along 19 streams (Figure 1) that were: i) distributed across the study

112 area; ii) located on first-to-fourth-order streams (Strahler 1952); and that iii) had at least a 120-m

113 reach surrounded by forest stands greater than 15 years in age. These criteria ensured that we

114 sampled a range of stream sizes but none so large that it prevented us from easily crossing the

115 stream to sample both sides. Also, by ensuring that riparian forest was greater than 15 years of

116 age, we were able to avoid stream banks that were inaccessible due to heavy accumulation of Draft for Review Only – Not for Distribution 6 | Hayes et al. 117 harvest debris (Jackson and Sturm 2002; personal observation). Eighteen streams were from

118 forests managed for timber; one was located on unmanaged land.

119

120 Sampling

121 Sampling overview.— We used a streambank survey approach modified from Raphael et

122 al. (2002) to characterize amphibian use across a gradient from the stream into adjacent uplands

123 (Figure 2a,b). At each stream, we sampled a series of 10 2-m wide plots (hereafter belts) arrayed

124 perpendicular to the stream (Figure 2b). We spaced belts along the stream at uniform intervals

125 ≥10 m but ≤30 m; the interval chosen at a particular stream depended on available area of stand

126 age ≥15 years. Minimum spacing between adjacent belts (10 m) was based on data suggesting

127 that western red-backed salamander and ensatina move over a scale of <10 m during their active

128 season (Gnaedinger and Reed 1948, Ovaska 1988, Ovaska and Gregory 1989, Davis 1998,

129 Maxcy and Richardson 2000), and data that appeared after the initiation of this study implied

130 similar movement scales for Dunn’s and Van Dyke’s salamanders (McIntyre 2004, McIntyre et

131 al. 2005; D. Olson, USDA Forest Service, personal communication). Sampling of adjacent belts

132 on a stream spanned up to 8 days, so this minimum spacing helped ensure independence among

133 belts.

134 Each belt extended from the wetted stream edge to the valley wall (i.e., top of the actively

135 eroded bank) on both sides of the stream (Figure 2a,b), hereafter banks. We also extended each

136 belt another 35 m above the valley wall on one side of the stream (Figure 2a,b), hereafter

137 uplands. If no clear valley wall break was present, the stream-adjacent tree line defined the bank-

138 upland boundary. We alternated the upland portion of belts between sides of the stream except

139 where an obstruction (e.g., cliff) prevented sampling the chosen side. To clarify habitat Draft for Review Only – Not for Distribution 7 | Hayes et al. 140 characterization with data obtained at the level of individual animals or microhabitat, hereafter

141 we term banks and uplands macrohabitats.

142 In the Pacific Northwest, sampling during spring and fall intervals, when surface

143 moisture is relatively high and freezing and elevated temperatures are rare, is believed to

144 increase the likelihood of detection for most terrestrial salamanders (e.g., Aubry et al. 1988,

145 Grialou et al. 2000, Aubry 2000),. We chose to sample in spring because this interval maintains

146 conditions conducive to sampling over a longer interval than the fall (Jones 1999). We sampled

147 diurnally (0800-2000 hr) over a similar seasonal interval in each of 2 years: 72 days (17 Apr-27

148 Jun) in 2001, 74 days (14 Apr-26 Jun) in 2003. We did not sample on days when air temperature

149 dropped to 0º C at any time over the previous 2 days.

150 Habitat sampling.— At each site, we initially characterized substrate moisture and

151 temperature, and substrate and vegetation composition along each belt. We recorded moisture

152 and temperature under easily moveable surface objects along the centerline of belts at 4 near

153 equidistant points on each bank and 12 equidistant points on each upland. We scored moisture

154 under easily moveable surface objects as wet, moist, or dry. A wet substrate had visible excess

155 water; a moist substrate lacked visible excess water, but paper toweling could absorb water from

156 it seconds after contact; and paper toweling was incapable of absorbing water from a dry

157 substrate. We measured temperature with digital or laser thermometers accurate to 0.1º C that

158 were calibrated both using a known standard and against each other for precision.

159 We characterized substrate by determining percent cover by type across the entire area of

160 each macrohabitat. We partitioned substrates into 3 basic categories: fines, inorganics, and

161 organics, and the latter 2 basic categories were further partitioned (Appendix 1). We treated fines

162 this way because we could not unambiguously categorize them as either organic or inorganic. Draft for Review Only – Not for Distribution 8 | Hayes et al. 163 Four of the 5 inorganic substrate categories (gravel, small cobble, large cobble, and boulder)

164 were based on a modified Wentworth (1922) scale; the fifth, bedrock, was consolidated rock too

165 large to gauge a particle size. We partitioned organic substrates into 7 categories (Appendix 1): 3

166 sizes of large woody debris (LWD; small, medium and large); non-LWD wood; leaf litter; and 2

167 categories of non-woody vegetation (moss and non-moss herbaceous vegetation). We separated

168 moss from non-moss vegetation because some terrestrial salamanders shelter in moss (Ovaska

169 and Gregory 1989, Dupuis et al. 1995). For each piece of LWD intersecting the belt, we

170 measured its midpoint diameter (across a belt) and its length. We also scored its decay class

171 based on Sollins (1982) and identified it to species when possible. We estimated vegetation

172 cover over the entire area of each macrohabitat to the nearest 5% for herb (including moss

173 separately), shrub, and canopy layers. We estimated percent cover for herb and shrub layers

174 visually. We estimated canopy cover using a 25-point grid moosehorn with a 6° viewing angle.

175 We defined the shrub layer as woody vegetation <5 m high plus sword fern (Polystichum

176 munitum), and recorded the dominant species in canopy and shrub layers.

177 To characterize potential climatic differences between years, we obtained monthly

178 precipitation and temperature data from 4 stations encompassing the area (Centralia [COOP No.

179 451276], Naselle [455744], Raymond [456914], and Toledo [458500]) over 6-month periods that

180 included our sampling interval as well as the preceding 3 months (January-March) each year.

181 Amphibian sampling.— After characterizing macrohabitat, we sampled each belt for

182 amphibians. We thoroughly searched belts by overturning all cover objects (e.g., rocks, LWD),

183 excavated substrate to a depth of 30 cm with a hand axe or pick unless solid rock prevented

184 reaching that depth, and dismantled LWD except where resistant sap- or heartwood prevented it. Draft for Review Only – Not for Distribution 9 | Hayes et al. 185 Captured animals were identified and released at the point of capture. For each animal,

186 we also recorded their macrohabitat location (bank or upland), distance to the stream, scored

187 whether or not we found them concealed, and collected data on microhabitat. Microhabitat data

188 included cover type (concealing an animal), substrate type, moisture (as described in habitat

189 sampling), and temperature of the substrate beneath the animal at capture. We scored cover and

190 substrate types using the same categories used to evaluate belt substrates. If wood associated

191 (acting as cover or substrate) with an animal was LWD (defined as minimum diam ≥10 cm and

192 length ≥ diam), we recorded diameter and decay class.

193 We followed Crother et al. (2008) for amphibian nomenclature except for Pseudacris

194 regilla, which we term Pacific chorus frog. To simplify salamander names, we designated

195 species by the first 2 letters of their respective scientific names: Dunn’s salamander (PLDU),

196 ensatina (ENES), Van Dyke’s salamander (PLVA), and western red-backed salamander (PLVE).

197 We handled animals under a permit that covers Washington Department of Fish and Wildlife

198 (WDFW) personnel conducting state work; per permit conditions, we followed the guidelines for

199 handling of live amphibians and reptiles in the field (Beaupre et al. 2004).

200

201 Analysis

202 Our habitat utilization analyses addressed 2 fundamental levels: among sites and within sites.

203 Sampling unique sites each year made between-year contrasts basic to analyses among sites.

204 Within-site analyses were paired comparisons between macrohabitats (banks versus uplands) in

205 which belts were the unit of comparison. For these comparisons, we assumed that the nesting of

206 belts within sites lacked significant autocorrelative effects. Low sample sizes for some species

207 disallowed some contrasts between either years or macrohabitats. For selected analyses, we also Draft for Review Only – Not for Distribution 10 | Hayes et al. 208 conducted comparisons with and without site 19 (Figure 1) because we observed obvious habitat

209 drying during sampling at this site that we thought might influence differences between years or

210 macrohabitats. Limited sample sizes led us to set α a priori at 0.1 for the site-level analyses; α

211 was 0.05 for all other analyses.

212 Background analyses.— Because we sampled different sites each year, we compared the

213 habitat characteristics of sites in 2001 versus 2003 to help determine whether those differences

214 might confound the ability to distinguish amphibian habitat use patterns. We first used Principle

215 Component Analysis (PCA) to explore differences in the physical and vegetation characteristics

216 between the sites sampled in each year. Specifically, PCA enables visual inspection of the

217 relationship among sites by examination of their relative position over defined gradients (i.e.,

218 principle components). To conduct the PCA, we calculated a Pearson correlation matrix among

219 11 variables characterizing physical structure and vegetation at each site (i.e., pooled belts); these

220 included percent cover of canopy, shrub, moss, mud, sand, gravel, small cobble, large cobble,

221 boulder, bedrock; and the numbers of pieces of LWD per square meter. Of the 11 variables, we

222 found that percent gravel and percent large cobble were each correlated (r > 0.7) with 3 other

223 variables, and hence were removed from the dataset. We arcsine-transformed the remaining 8

224 percentage-based variables and square root-transformed the LWD tally variable to normalize

225 their distributions. All transformed variables approximated normal distributions better than their

226 untransformed counterparts based on visual inspection of probability plots, and no transformed

227 variable had a distribution different from normal as determined with a Lilliefors tests (P >

228 0.400). We then obtained non-rotated and varimax-rotated principle components from the

229 correlation matrix of the 9 transformed variables (McGarigal et al. 2000). We chose varimax-

230 rotated over simple PCA for visualization because it enhanced interpretation of the first 2 Draft for Review Only – Not for Distribution 11 | Hayes et al. 231 components, i.e., provided fewer factors with larger loadings. To ensure no differences were

232 concealed, we also performed univariate contrasts using Mann-Whitney tests for each habitat

233 variable (pooled across all belts at each site) between years.

234 We characterized regional-level precipitation and temperature data to identify general

235 differences between sample years. For the 4 stations from which we obtained monthly climatic

236 data, we descriptively compared the 3-month intervals prior to and during sampling in each year.

237 To identify potential differences between the set of sites sampled each year, we collectively

238 compared data collected from all points over the seasonal interval each year with a Mann-

239 Whitney test for temperature and moisture. We compared banks and uplands separately because

240 of the expected differences between banks and uplands in moisture (Robert and Spies 1998) or

241 temperature (Brosofske et al. 1997).

242 To thoroughly explore an amphibian-wood association, we characterized wood in several

243 ways. These included the density of LWD, the volume and density of functional LWD, and the

244 volume and density of LWD >50 cm in diameter (hereafter LWD50). We quantified density as

245 pieces per square meter and volume as cubic meters per square meter, and functional LWD

246 referred to LWD in decay classes 2, 3, or 4 (Sollins 1982). We included the functional wood

247 category because terrestrial salamanders may make greater use of LWD in intermediate decay

248 classes (e.g., Aubry et al. 1988, Aubry and Hall 1991), and larger diameter LWD was of special

249 interest as a potentially limiting resource in managed forests (Olson et al. 2006).

250 We used Wilcoxon Signed Rank tests for paired contrasts to compare bank versus upland

251 habitat characteristics across all sites, by year because we expected that banks, which are

252 equivalent to riparian areas as defined herein, differed from uplands (Olson et al. 2007). We also Draft for Review Only – Not for Distribution 12 | Hayes et al. 253 compared moisture using a Wilcoxon Signed Rank test and temperature data using a paired t-test

254 between banks and uplands across all sites by years.

255 We summarized terrestrial salamander occupancy data at the site, belt, and macrohabitat

256 levels each year. We compared site-level occupancy between years for each species using

257 Fisher’s Exact Tests, belt-level occupancy by species among sites each year using Mann-

258 Whitney Tests; and occupancy between banks and uplands in 2001 using Wilcoxon Signed Rank

259 Tests.

260 Terrestrial salamander habitat utilization.— We compared habitat utilization in 2 ways:

261 distribution of animals relative to the stream and animal use of microhabitat. We first compared

262 salamander density between banks and uplands by species for each year using unpaired t-tests.

263 Due to variable bank lengths (Figure 2a), we also compared animal densities by species in 6-m

264 intervals along belts (ignoring macrohabitats) using a single factor ANOVA for each year to

265 determine if density changed as a function of absolute distance to the stream. We chose the 6-m

266 length to ensure adequate captures (non-zero data). This resulted in 6 6-m intervals for each belt;

267 we excluded data >36 m from each stream because only a fraction of belts exceeded this distance

268 at each site. Data used in this analysis were the means of belt densities at each site. For each

269 species, we excluded those sites where the species was not detected (no detections in any belt).

270 To address microhabitat use, we compared the proportions of microhabitats used by

271 substrate category between the sets of sites sampled each year using a Kolmogorov-Smirnov test.

272 To provide some comparability to previous studies indicating that Dunn’s salamanders were

273 associated with inorganic substrates (Fitch 1936, Dumas 1956, Bury and Corn 1991, Wilkins and

274 Peterson 2000), we used the 3 basic substrate category groupings: fines, inorganic, and organic

275 (Appendix 1). To address wood use, we re-did this analysis after repartitioning the organic Draft for Review Only – Not for Distribution 13 | Hayes et al. 276 category into large wood (LWD >50 cm diam), all remaining wood, and non-wood; to keep the

277 analysis orthogonal, non-wood included fines, all inorganic substrates, and non-wood organic

278 substrates (Appendix 1). We performed separate analyses for each type of cover and substrate

279 (where the animal sat). Cover analyses excluded a small percentage (<5%) of unconcealed

280 animals.

281 Use versus available habitat.— We contrasted amphibian use versus availability of

282 habitat in 2 ways, one for temperature and moisture, and the other for wood and other substrates.

283 We compared temperature and moisture characteristics at the location of each individual (used

284 habitat) versus conditions in the surrounding macrohabitat (i.e., banks or uplands) in which each

285 salamander was found (available habitat). We did this analysis for each year for each species

286 using paired t-tests (temperature) or Wilcoxon Signed Rank tests (moisture). We obtained use

287 and availability temperature and moisture data within 1 hour of each other. To examine use

288 versus availability of wood, we conducted Spearman rank correlations between the proportion of

289 terrestrial salamanders found in wood at each site and the amount of available wood at that site.

290 We considered all measures of wood previously described. We were able to examine these

291 relationships only for PLDU, PLVA, and PLVE on banks in 2001. For species positively

292 associated with some measure of wood, we also determined whether a potential relationship

293 existed to alternative (inorganic) substrates with a second correlation analysis. Dunn’s

294 salamanders may be associated with rocks (Fitch 1936, Dumas 1956, Bury and Corn 1991, Bury

295 et al. 1991, Wilkins and Peterson 2000), but these studies show associations with different

296 particle sizes (for example, compare Bury et al. 1991 and Wilkins and Peterson 2000). Thus, we

297 evaluated 2 particle size ranges for inorganic substrates: one that reflects the association with

298 pebbles (= our gravel; see Appendix 1) reported by Wilkins and Peterson (2000) and the other, Draft for Review Only – Not for Distribution 14 | Hayes et al. 299 the broad range of inorganic cover reported by Bury et al. (1991), which we interpreted to

300 represent gravel to boulder.

301

302 RESULTS

303 Background

304 Overstory vegetation on banks was more diverse than overstory in uplands. Uplands were

305 dominated by western hemlock (Tsuga heterophylla; 13 sites) and Douglas-fir (Pseudotsuga

306 menziesii; 6 sites). Besides western hemlock and Douglas-fir, banks also contained a relatively

307 high proportion of red alder (Alnus rubra), Sitka spruce (Picea sitchensis) and western red cedar

308 (Thuja plicata). Huckleberries (Vaccinium spp.), sword fern, and salal (Gaultheria shallon)

309 dominated the shrub layer in uplands, but salmonberry (Rubus spectabilis), sword fern and salal

310 dominated the shrub layer on banks.

311 We sampled 15,901 m2 of terrestrial habitat during streambank surveys: 10,961 m2 in

312 2001, and 4,939 m2 in 2003. The 10-belt area at each site averaged 836 m2 (SD = 251 m2),

313 variation resulting entirely from differences in bank lengths (Figure 2a); and areas of sites

314 sampled in 2001 did not differ from those sampled in 2003 (Z17 = -1.213, P = 0.229). We also

315 found no obvious relationship among sites by year after examining their relative position on PCA

316 gradients (Figure 3). The first 2 principal components explained 50.8% of the variation. The first

317 principal component scores increased as percent cover of small cobble, bedrock, and boulders

318 increased (factor loadings of 0.86, 0.83, and 0.74, respectively) and decreased with increasing

319 amounts of canopy (factor loading = -0.60). The second principal component increased with

320 increasing amounts of sand (factor loading 0.91) and decreased with increasing amounts of moss

321 (factor loading -0.84). Except for the percentage of moss cover, we also found no differences in Draft for Review Only – Not for Distribution 15 | Hayes et al. 322 physical or vegetation characteristics between sample years as determined by univariate

323 analyses. The percentage of moss cover averaged almost twice as high on sites from 2001 ( x =

324 45.8%, SD = 20.7) as those sampled in 2003 ( x = 24.8%, SD = 14.0; Z17 = -1.792, P = 0.079).

325 We recorded greater homogeneity in all other variables (Z17 ≤ -1.312, P ≥ 0.179 for all

326 combinations), and the only unmanaged site fell within the range of values for managed sites

327 sampled each year.

328 Though we found no obvious differences in habitat characteristics at sites sampled

329 between years, the climate differed dramatically between years. Over January-March, the 3

330 months before we sampled, precipitation in 2003 was twice that in 2001 (Figure 4). We found

331 this pattern almost reversed, though at a lower absolute level, during April-June, at which time

332 precipitation in 2003 was 9-32% less than in 2001, variation reflecting differences among

333 stations (Figure 4). We recorded higher temperatures in 2003 than in 2001 on banks and uplands,

334 inclusive of site 19 or not (banks: with site 19: t365 = -5.496, without site 19: t347 = -3.883;

335 uplands: with site 19: t552 = -4.380, without site 19: t524 = -2.522; P ≤ 0.012 for all 4). We also

336 found greater moisture on banks and uplands in 2003 versus 2001 (banks: Z382 = -2.569; uplands:

337 Z574 = -1.950, P ≤ 0.010 for both), but this difference vanished when we removed site 19 (banks:

338 Z362 = -0.445; uplands: Z544 = -1.210; P ≥ 0.226 for both).

339 Though we found no differences in the amount of wood between sites sampled each year,

340 densities or volumes of the various measures of wood varied between 3-to-4-fold among sites.

341 Based on measures of either density or volume, we also generally found more wood on banks

342 than uplands.

343 We also found differences in soil moisture and temperature between banks and uplands.

344 Banks were wetter than uplands in both years (2001: Z139 = -5.044; 2003: Z49 = -3.784; P < 0.001 Draft for Review Only – Not for Distribution 16 | Hayes et al.

345 for both), a difference in 2003 unaffected by excluding site 19 (Z39 = -2.850, P = 0.004). Bank

346 temperatures ( x = 10.9º C, SD = 2.8º C) were lower than upland temperatures ( x = 11.6º C,

347 SD = 3.2º C) in 2001 (t131 = -6.732, P ≤ 0.001), but banks ( x = 12.9º C, SD = 3.4º C) did not

348 differ from uplands ( x = 13.1º C, SD = 3.6º C) in 2003, regardless of whether site 19 was

349 excluded (all sites: t49 = -0.629; without site 19:t39 = -0.330; P ≥ 0.532 for both).

350

351 Amphibian Sampling

352 We detected 824 amphibians representing the same 12 species during streambank surveys

353 over both years (Table 1): 466 in 2001, 358 in 2003. Terrestrial salamanders comprised just over

354 75% (n = 353) of all amphibians found in 2001, but over 91% (n = 327) in 2003. We sampled

355 only about one-third as many sites in 2003 as in 2001, but found only 7% fewer terrestrial

356 salamanders that year.

357 Overall, terrestrial salamander densities were greater in 2003 than in 2001 (Z17 = -2.867,

358 P = 0.004). However, we found contrasting species-specific patterns. We recorded ENES and

359 PLVE, respectively, at more belts per site (Z17 = -2.573, P = 0.010; Z17 = -2.541, P = 0.011) and

360 at higher densities (Z17 = -2.226, P = 0.026: Z17 = -2.037, P = 0.042) in 2003, but found the

361 opposite pattern for PLVA, which occurred at more belts per site (Z17 = -2.083, P = 0.037) and

362 higher densities (Z17 = -1.958, P = 0.050) in 2001. We found no differences between years in

363 belt-level occupancy or density for PLDU (occupancy: Z17 = -0.141, P = 0.888; density: Z17 =

364 -0.093, P = 0.926). We also found 8 non-terrestrial amphibian species at low densities (Table 1);

365 we do not address these species further.

366 Terrestrial salamander habitat utilization.— We found marked differences in patterns of

367 macrohabitat use among the 4 terrestrial salamander species. Regardless of year, PLDU and Draft for Review Only – Not for Distribution 17 | Hayes et al. 368 PLVA were infrequent in uplands, while ENES and PLVE were more evenly spread over banks

369 and uplands (Table 2). Even though ENES was more common than either PLDU or PLVA, its

370 belt-level occupancy revealed a pattern somewhat opposite to these 2 species; ENES was 3 times

371 more abundant in uplands than banks (Table 2). The most common species, PLVE, had a year-

372 specific pattern. In 2001, PLVE occurred about 20% more frequently in uplands than banks, but

373 in 2003 we found no difference in occupancy between banks and uplands (Table 2). With a few

374 caveats, we recorded similar patterns based on density. Densities of PLDU in both years and

375 PLVA in 2001 were greater along banks than in uplands (Table 3). In contrast, in both years we

376 found higher densities of ENES in uplands than on banks, and no differences between bank and

377 upland densities of PLVE. Further, except for PLVA, we recorded higher densities of each

378 species in all macrohabitats in 2003 (the wet year) than 2001 (the dry year).

379 We observed similar distributional patterns when we considered absolute distance to the

380 stream based on 6-m segments. In 2001, PLDU and PLVA displayed differences in densities

381 with distance from the stream (PLDU: F5 = 17.2; PLVA: F5 = 19.5, P < 0.001 for both; Figure

382 5). For both species, we found greater densities between the stream-adjacent and all other 6-m

383 intervals, but no differences between remaining interval combinations (Figure 5). In contrast, we

384 found no differences in density with distance from the stream in 2001 for either ENES (F5 = 1.0,

385 P = 0.425) or PLVE (H5 = 1.8, P = 0.129; Figure 5), and we found the identical pattern in 2003

386 (ENES: F5 = 1.8, P = 0.153) or PLVE (F5 = 0.3, P = 0.897).

387 Nearly all terrestrial salamanders were concealed under cover, the frequency of which

388 ranged from 94% for ENES in 2003 to 100% for PLDU in 2003 and PLVA in both years. We

389 found no differences in the proportion concealed by cover among species, regardless of year

2 390 (χ 3 ≤ 2.6, P ≥ 0.464 for both years). Draft for Review Only – Not for Distribution 18 | Hayes et al. 391 We found selected differences in the patterns of microhabitat use among the 4 species for

392 moisture, temperature, and substrate. PLVE used moister microhabitats on banks than in uplands

393 in 2003 (Z217 = -3.007, P = 0.003), even when site 19 was removed (Z211 = -2.903, P = 0.004).

394 However, we did not observe that difference for PLVE in 2001 (Z219 = -1.266, P = 0.206) or

395 ENES in either year (2001: Z49 = 0, P > 0.999; 2003: Z51 = -0.953, P = 0.341). Further, PLDU

396 used wetter microhabitats on banks in 2001, even when site 19 was removed (Z61 = -3.345, P <

397 0.001 for both alternatives). ENES used wetter microhabitats in uplands in 2001 (Z82 = -1.989, P

398 = 0.047), but this difference disappeared when site 19 was removed (Z79 = -0.953, P = 0.076).

399 We found no differences in the moisture level of microhabitats used by PLVE between years

400 either in banks (Z196 = -0.521, P = 0.603) or uplands (Z240 = -1.569, P = 0.117), even if we

401 removed site 19 (banks: Z194 = -0.455, P = 0.649; uplands: Z235 = -1.328, P = 0.184). In 2001,

402 PLDU and PLVA each used microhabitats on banks that were wetter than those used by PLVE

403 (PLDU: Z124 = -3.065, P = 0.002; PLVA: Z107 = -4.067, P < 0.001). We did not see this pattern

404 for PLDU in 2003, even when site 19 was removed (Z102 = -1.862, P = 0.118 for both

405 alternatives). PLDU and PLVA used microhabitats with similar moisture levels in banks in 2001

406 (Z48 = -0.729, P = 0.466). In 2003, PLVE utilized wetter microhabitats than ENES in both banks

407 (Z100 = -2.801, P = 0.005) and uplands (Z168 = -2.848, P = 0.004). When we excluded site 19, the

408 difference in uplands remained (Z160 = -3.044, P = 0.002), but the difference in banks marginally

409 disappeared (Z96 = -1.888, P = 0.059).

410 We found selected differences in the temperature and moisture of microhabitats utilized

411 by the 2 widespread terrestrial salamander species between banks and uplands. ENES used sites

412 having similar temperature and moisture in banks versus uplands in 2001 (t49 = 0.588, P = 0.559)

413 and 2003 (t51 = 0.032, P = 0.974), as did PLVE in 2001 (t218 = -0.088, P = 0.930; Table 4). Draft for Review Only – Not for Distribution 19 | Hayes et al.

414 However, we recorded PLVE in cooler microhabitats in uplands versus banks in 2003 (t217 =

415 3.767, P < 0.001; Table 4).

416 We found species-, habitat-, and year-specific variation in terrestrial salamander use of

417 both cover and substrates. Analyses based on groupings by fines, inorganics and organics

418 revealed that the 4 species utilized different cover categories in both years along banks (2001:

2 2 419 χ 6 = 19.2, P = 0.004; 2003: χ 6 = 23.8, P = 0.001; Table 5), but parallel analysis for ENES and

2 2 420 PLVE in uplands revealed no differences (2001: χ 4 = 2.8, P = 0.590; 2003: χ 4 = 4.3,

421 P = 0.362). About two-thirds of all ENES, PLDU, and PLVE were found under organic cover in

422 both years. Only PLDU and PLVE used inorganic cover with some regularity, and only PLDU

423 used that cover category over a third of the time (Table 5). All species infrequently used fines as

424 cover. Regardless of year, PLVE used organic cover more often in uplands than along banks

2 2 425 (2001: χ 2 = 17.0, P < 0.001; 2003: χ 2 = 12.2, P = 0.002); we found the same pattern for PLDU

2 2 426 in 2003 (χ 2 = 7.3, P = 0.026), but not in 2001 (χ 2 = 3.8, P = 0.149). We did not observe this

2 2 427 pattern in ENES in either year (2001: χ 2 = 1.1, P = 0.586; 2003: χ 1 < 0.1, P = 0.947). All

428 species used organics less frequently and fines more frequently as substrate than for cover, yet,

429 we identified several patterns similar to those observed for cover. Notably, PLVA used organic

430 substrates most frequently; PLDU and PLVE regularly used inorganic substrates; only PLDU

431 used inorganic substrates more than a third of the time, though here, the pattern was not

432 restricted to banks; and PLVE utilized organic substrates more often in uplands than along banks

2 2 433 (2001: χ 2 = 6.2, P = 0.045; 2003: χ 2 = 10.4, P = 0.005).

434 All terrestrial salamander species made some use of wood as cover, but we also found

435 species-, habitat-, and year-specific variation. We found significant differences in the proportion

2 436 of animals using wood as cover among species on banks in both years (2001: χ 6 = 18.1, Draft for Review Only – Not for Distribution 20 | Hayes et al.

2 2 437 P = 0.006; 2003: χ 6 = 16.8, P = 0.010; Table 6), but not in uplands (2001: χ 4 = 1.3, P = 0.869;

2 438 2003: χ 4 = 3.0, P = 0.551). Only PLVA had large wood use >25%. Overall use of wood as cover

439 varied with macrohabitat and year; 33% to 70% for ENES, 17% to 60% for PLDU, and 40% to

440 58% for PLVE. Moreover, use of large wood as cover was never >15% for ENES or PLVE, and

441 never >20% for PLDU. Except for ENES in 2001, overall wood use as cover was greater in

442 uplands than on banks across all species, but this pattern was significant only for PLVE in 2003

443 (Fisher’s Exact test: P = 0.008). We also found no differences in overall wood use as cover

444 between years for any species except for PLDU on banks, for which its use as cover was greater

445 in 2001 than 2003 (Fisher’s Exact test: P = 0.031). In contrast, except for the lone PLVA in

446 2003, all species used non-wood materials more often as substrate than for cover. Excluding the

447 lone observation in 2003, PLVA was the only species for which overall wood use as substrate

448 exceeded 50%. As with cover, significant differences in wood utilization as substrate were found

2 2 449 among species on banks in both years (2001: χ 6 = 16.4, P = 0.012; 2003: χ 6 = 26.4, P < 0.001),

2 2 450 but not in uplands (2001: χ 6 = 1.8, P = 0.936; 2003: χ 4 = 1.3, P = 0.851; Table 6).

451 Habitat use versus availability.— We found distinctive patterns among species in use

452 versus availability relative to substrate moisture and temperature. On banks, PLDU and PLVA

453 used wetter sites than were available in 2001, but we found no difference in the moisture of

454 available versus used habitat for PLDU in 2003 (Table 7). In uplands, we also found that PLVE

455 used wetter sites than were available 2001, but not in 2003.

456 We found that PLDU on banks used cooler sites than were available in 2001, but we did

457 not find this pattern either for PLVA in 2001 or for PLDU in 2003 (Table 7). In 2001, we also

458 found PLVE on banks at warmer sites than were available (Table 7). However, in uplands, we

459 also found that ENES and PLVE used cooler sites than were available in 2001. Interestingly, this Draft for Review Only – Not for Distribution 21 | Hayes et al. 460 pattern was reversed for ENES in 2003, i.e., ENES used warmer sites than were available; we

461 detected no difference in the temperature of microhabitat used versus available for PLVE in

462 2003.

463 We found a negative relationship between the proportion of PLDU using wood as cover

464 and the density of pieces of LWD >50 cm in diameter on banks in 2001 (Table 8). The

465 relationship between wood used by PLDU and available wood tended to be negative regardless

466 of how wood was measured. In contrast, PLVA tended to show a positive relationship between

467 the proportion of animals using wood cover and various categories of wood, though this

468 relationship was significant only for LWD density. We also found no obvious relationship with

469 inorganic substrates for either PLDU or PLVA, regardless of whether we measured gravel or

470 gravel-to-boulder categories (ρ ≤ 0.123). We also found no relationship between PLVE and

471 wood.

472

473 DISCUSSION

474 Our findings generally agree with available knowledge on the ecology and habitat use of these 4

475 terrestrial salamanders. In western Oregon and Washington, ENES and PLVE are by far the most

476 often encountered terrestrial salamander species (e.g., McComb et al. 1993, Aubry 2000, Grialou

477 et al. 2000, Butts and McComb 2000) and both occur over a wide range of forested habitats that

478 includes uplands and stream margins (Aubry and Hall 1991, Gilbert and Allwine 1991). By

479 contrast, PLDU and PLVA are less frequently encountered (Gilbert and Allwine 1991, Wilson et

480 al. 1995, Vesely and McComb 2002) and PLDU has been found close to streams more frequently

481 than in uplands in Oregon (Gomez and Anthony 1996, Vesely and McComb 2002). Our work

482 showed that PLVA also exhibits a stream-proximate distributional pattern, particularly in a dry Draft for Review Only – Not for Distribution 22 | Hayes et al. 483 year. Previous work, focusing on its hydrophile status, addressed seeps (McIntyre et al. 2005,

484 Crisafulli et al. 2005) or only stream-proximate habitat (McIntyre 2004).

485 The moisture selection pattern we found for terrestrial salamanders is consistent with the

486 idea that PLDU and PLVA tend to be relatively hydrophilic. In contrast to ENES and PLVE,

487 PLDU and PLVA were located mostly in wetter macrohabitat (banks), and selected wetter

488 microhabitats on banks in the dry year. This is similar to the observation of Dumas (1956), who

489 found PLDU used a wetter substratum at lower humidities. Since the most prominent feature of

490 both PLDU and PLVA macrohabitat distributions was the use of streambanks (versus uplands),

491 we suggest that this pattern reflects the greater reliability of banks in providing suitably cool, wet

492 habitat. Moreover, in the drier macrohabitat (uplands) PLVE, the more mesic-adapted of the 2

493 widespread species (Grialou et al. 2000), was the only species that selected wetter microhabitats

494 in both years; a pattern that tended to be stronger in the drier year. We also found selected

495 parallels with temperature patterns in the dry year. Namely, PLDU selected cooler temperatures

496 than available on banks, and both ENES and PLVE selected cooler temperatures than available

497 in uplands in 2001. In addition, PLVE also selected cooler temperatures than available in uplands

498 during the wet year, but selection appeared to be weaker than in the drier year. The only

499 inconsistent pattern we found was related to fact that PLVE used cooler temperatures than were

500 available on banks in the dry year, which was also the cooler year.

501 Our findings regarding substrate utilization are only indirectly comparable to previous

502 work due to differences in sampling methods, constraints associated either with site selection or

503 the types of substrate considered (usually some type of LWD), association of animals with

504 nearby substrates (rather than the substrate on or in which animals were found) or some

505 combination of these. For example, based on area-constrained ground searches in which LWD Draft for Review Only – Not for Distribution 23 | Hayes et al. 506 was dismantled, Butts and McComb (2000) stated that most amphibians captured, 89% of which

507 were ENES, were found in or under the bark of decay class 3 or 4 logs or log fragments, adding

508 that they also often occurred under loose bark, moss, or organic litter. We found ENES on or in

509 all the substrate types they discuss, and their statement also implies that ENES used a high

510 proportion of organic substrates, which is similar to what we observed. However, the context of

511 the use observed by Butts and McComb (2000), for example whether or not it was constrained

512 by lack of alternative substrates, is not known. Comparability of substrate-based data for all the

513 other species is equally limited. Our findings that PLDU was frequently associated with

514 inorganic substrates (Herrington 1988, Corn and Bury 1991, Wilkins and Peterson 2000) and that

515 PLVE uses diverse substrates (Aubry and Hall 1991, Gilbert and Allwine 1991) are generally

516 consistent with previous studies. However, the association of PLDU with inorganic

517 microhabitats, from this and previous studies, may simply be a consequence of its proximity to

518 the stream, where stream processes (fluvial or mechanical) expose inorganic substrates with

519 greater frequency than in uplands. As discussed previously, we believe that PLDU use

520 streambanks because they tend to be wetter and cooler than uplands. Previous habitat work on

521 PLVA, limited to high-gradient aquatic habitats (McIntrye 2004, McIntyre et al. 2005), showed

522 that PLVA occurrence increased on certain inorganic substrates, namely small cobble, small

523 gravel, and bedrock. This pattern of substrate use is dramatically different from our data, which

524 showed a strong association between PLVA and organic substrates, especially large wood.

525 The value of this study lies in its demonstration that species-specific substrate utilization

526 can change between years likely because of changes in climatic conditions. For example, though

527 we found no PLDU in large wood during the wet year (2003), the overall use of wood (large and

528 other wood) by PLDU roughly doubled during the dry year. Moreover, PLDU preferred wetter, Draft for Review Only – Not for Distribution 24 | Hayes et al. 529 cooler microhabitats than were available during the dry, but not the wet year. These habitat use

530 patterns make sense in light of the large difference in precipitation between the 2 years, and the

531 idea that in dry years, wood substrates provide this species moisture and thermal advantages over

532 mineral substrates (Heatwole 1962, Bunnell et al. 2002). Use of wood may also be related to

533 habitat needs of other life history stages. For example, one of the only 2 nests of PLDU and 7 of

534 the 8 nests of PLVA known to science were found in decayed logs (Jones 1989, Nauman et al.

535 1999, Blessing et al. 1999; M. Hayes, unpublished data). Our data also suggest that even species

536 thought to be relatively insensitive to drying conditions use wetter, cooler microhabitats than are

537 generally available during relatively warmer or drier conditions. Under the most extreme

538 conditions examined in this study (uplands in the dry year), we found that both ENES and PLVE

539 used cooler microhabitats and PLVE also used wetter microhabitats than were available.

540 Changes in habitat utilization by terrestrial salamanders in response to local weather

541 conditions underscore the need to better understand how these species will fare under climate

542 change scenarios. The Pacific Northwest is expected to experience increased climatic variability

543 under most climate change scenarios (Cayan 1996, Cayan et al. 2001). Refuges in large LWD

544 may become progressively more important if increases in variability translate into periods of

545 extreme summer temperatures (i.e., ≥30° C). We recorded one such extreme event in the interval

546 29 July-4 August 2003, where daily high air temperatures beneath forest canopy exceeded ≥32°

547 C on at least 2 days. During this event we discovered many PLDU and PLVE immersing

548 themselves in the streams (Eaton-Mordas et al. 2004) presumably as a way of avoiding lethal

549 temperatures (Dumas 1956).

550 We are somewhat perplexed by the results related to PLVA. We recorded this widely

551 acknowledged hydrophilic species at 9 of 14 sites during the drier year, but at only 1 of 5 sites Draft for Review Only – Not for Distribution 25 | Hayes et al. 552 during the wetter year. In addition, densities of PLVA were relatively high in years when the

553 densities of the other 3 species were low. Perhaps the most surprising finding was the relatively

554 high proportion of sites at which we found PLVA across the managed landscape. In a study on

555 the Olympic Pennisula, Raphael et al. (2002) recorded PLVA solely in old-growth stands or old-

556 growth riparian management areas (buffers) even though they surveyed 42 second-growth

557 managed stands with diverse buffer configurations. We suspect that PLVA is particularly

558 sensitive to climatic conditions and recommend that future work on PLVA carefully consider

559 climate in comparative study designs.

560

561 MANAGEMENT IMPLICATIONS

562 Our work helps validate findings from previous work in western Oregon revealing that PLDU is

563 stream-associated despite the generally high precipitation levels found in their southwest

564 Washington range. Hence, strategies to protect this species through streambank management

565 (Vesely and McComb 2002) can be generalized to the species' entire geographic range. From a

566 wood utilization perspective, our data indicates that PLDU may be less sensitive to losses of

567 LWD than other terrestrial species such as PLVA, assuming streambank buffering is adequate

568 (see Vesely and McComb 2002 for a discussion). However, the fact that they selected cooler and

569 wetter microhabitat and were more closely associated with wood during the dry year suggests

570 that wood may be important under certain circumstances. More important, these findings

571 underscore the need to address habitat use across a range of climatic conditions. This may be

572 especially important under climate change scenarios, which suggest that the Pacific Northwest

573 will experience increased climatic variability (Cayan 1996, Cayan et al. 2001). Draft for Review Only – Not for Distribution 26 | Hayes et al. 574 In contrast, we can provide little management guidance for PLVA due to our lack of

575 understanding of seemingly conflicting observations. The between-year differences we observed

576 seem counter-intuitive. Consistent with our expectations, PLVA selected relatively wet

577 microhabitats and was frequently associated with wood during the dry year. However, as one of

578 the most hydrophilic of Pacific Northwest terrestrial salamanders (McIntrye et al. 2006), we

579 found very few animals during the wet year. Similar to PLDU, PLVA seems to be a near-stream-

580 associated species, but we do not know if this pattern changes during wet years. Further, we

581 cannot readily explain why we found PLVA at over half the managed sites we surveyed, whereas

582 Raphael et al. (2002) did not detect PLVA at any managed sites except those with an old-growth

583 component. This conflict may have an ecological or a phylogenetic basis or both. In any case,

584 PLVA habitat use and its potential association with large wood, an increasingly atypical feature

585 of managed forests, surely requires investigation.

586 Our data also suggest that LWD may provide a critical moisture or temperature refuge in

587 upland habitats in dry years for the 2 widespread species, ENES and PLVE. In managed

588 landscapes in Washington, timber harvest occurs primarily in uplands and leave tree

589 requirements (e.g., wildlife trees, shelterwood, snags) are often met in riparian buffers. But

590 downed wood of quality immediately useable as habitat or live wood sources that contribute

591 wood that can ultimately serve as habitat in uplands are limited. However, the importance of

592 such refuges remains unclear.

593

594 ACKNOWLEDGMENTS

595 The Washington Department of Natural Resources, through the Forest and Fish Adaptive

596 Management Program involving the Cooperative Monitoring Evaluation and Research Draft for Review Only – Not for Distribution 27 | Hayes et al. 597 Committee, the Landscape and Wildlife Advisory Group, and the Amphibian Research

598 Consortium helped fund this study. The Campbell Group, Longview Timberlands, the

599 Washington Department of Natural Resources, and Weyerhaeuser generously provided access to

600 their lands; provided substantial logistic assistance in terms of maps, selected electronic site data,

601 weather data, and support on safety issues. L. Jones, the originator of the streambank survey

602 approach used in this study, provided insights into its application. K. Aubry, S. Butts, A. Kroll,

603 and C. Mendoza provided valuable suggestions on various drafts of this manuscript. D. Olson

604 generously shared unpublished movement data on Dunn’s salamander. R. O’Donnell helped

605 develop Figure 1. R. Baker, B. Barmettler, P. Barmettler, A. Basey, E. Dirks, D. Dugger, A.

606 Gucker, M. Hunter, D. Jacobson, M. Kluber, S. Lampman, A. Mason, D. Morrow, A Palmeri-

607 Miles, A Pauley, D. Peebles, R. Price, C. Sato, and J. Wisniewski helped collect or enter data.

608

609 LITERATURE CITED

610 Aubry, K. B. 2000. Amphibians in managed, second-growth Douglas-fir forests. Journal of

611 Wildlife Management 64: 1041-1052.

612 Aubry, K. B., and P. A. Hall. 1991. Terrestrial amphibian communities in the southern

613 Washington Cascade Range. Pages 327-338 in L. F. Ruggiero, K. B. Aubry, A. B. Carey,

614 and M. H. Huff, technical coordinators. Wildlife and vegetation of unmanaged Douglas-

615 fir forests. U.S. Forest Service, General Technical Report PNW-GTR-285.

616 Aubry, K. B., L. L. C. Jones, and P. A. Hall. 1988. Use of woody debris by plethodontid

617 salamanders in Douglas-fir forests in Washington. Pages 32-37 in R. C. Szaro, K. E.

618 Severson, and D. R. Patton, technical coordinators. Proceedings of the symposium on

619 management of amphibians, reptiles, and small mammals in North America. U.S. Forest Draft for Review Only – Not for Distribution 28 | Hayes et al. 620 Service, General Technical Report RM-166.

621 Beaupre, S. J., E. R. Jacobson, H. B. Lillywhite, and K. Zamudio. 2004. Guidelines for use of

622 live amphibians and reptiles in field and laboratory research, Second edition.

623 Herpetological Animal Care and Use Committee (HACC), American Society of

624 Ichthyologists and Herpetologists.

625 Blessing, B. J., E. P. Phenix, L. L. C. Jones, and M. G. Raphael. 1999. Nests of Van Dyke’s

626 (Plethodon vandykei) from the Olympic Peninsula, Washington. Northwestern Naturalist

627 80: 77-81.

628 Brosofske, K. D., J. Chen, R. J. Naiman, and J. F. Franklin. 1997. Harvesting effects on

629 microclimatic gradients from small streams to uplands in western Washington.

630 Ecological Applications 7:1188-1200.

631 Bull, E. 2002. The value of coarse woody debris to vertebrates in the Pacific Northwest. Pages

632 171-178 in W. F. Laudenslayer, Jr., P. J. Shea, B. E. Valentine, C. P. Weatherspoon, and

633 T. E. Lisle, technical coordinators. Proceedings on the ecology and management of dead

634 wood in western forests. U.S. Forest Service, General Technical Report PSW-GTR-181.

635 Bunnell, F. L., L. L. Kremsater, and R. W. Wells. 1997. Likely consequences of forest

636 management on terrestrial, forest-dwelling vertebrates in Oregon. Report M-7 of the

637 Centre for Applied Conservation Biology, Department of Forest Sciences, University of

638 British Columbia, British Columbia, Vancouver, Canada.

639 Bunnell, F. L., L. L. Kremsater, and R. W. Wells. 1999. Managing to sustain vertebrate richness

640 in forests of the Pacific Northwest: Relationships within stands. Environmental Review

641 7:97-146.

642 Bunnell, F. L., I. Houde, B. Johnston, and E. Wind. 2002. How dead trees sustain live organisms Draft for Review Only – Not for Distribution 29 | Hayes et al. 643 in western forests. Pages 291-318 in W. F. Laudenslayer, Jr., P. J. Shea, B. E. Valentine,

644 C. P. Weatherspoon, and T. E. Lisle, technical coordinators. Proceedings on the ecology

645 and management of dead wood in western forests. U.S. Forest Service, General Technical

646 Report PSW-GTR-181.

647 Bury, R. B., and P. C. Corn. 1991. Sampling methods for amphibians in streams in the Pacific

648 Northwest. Pages 1-29 in A. B. Carey, and L. F. Ruggiero, technical editors. Wildlife-

649 habitat relationships: Sampling procedures for Pacific Northwest vertebrates. U.S. Forest

650 Service, General Technical Report PNW-GTR-275.

651 Bury, R. B., P. C. Corn, K. B. Aubry, F. F. Gilbert, and L. L. C. Jones. 1991. Aquatic amphibian

652 communities in Oregon and Washington. Pages 353-362 in L. F. Ruggiero, K. B. Aubry,

653 A. B. Carey, and M. H. Huff, technical coordinators. Wildlife and vegetation of

654 unmanaged Douglas-fir forests. U.S. Forest Service, General Technical Report PNW-

655 GTR-285.

656 Butts, S. R., and W. C. McComb. 2000. Associations of forest floor vertebrates with coarse

657 woody debris in managed forests of western Oregon. Journal of Wildlife Management

658 64: 95-104.

659 Cayan, D. R. 1996. Interannual climate variability and snowpack in the western United States.

660 Journal of Climate 9: 928-948.

661 Cayan, D. R., S. A. Kammerdiener, M. D. Dettinger, J. M. Caprio, and D. H. Peterson. 2001.

662 Changes in the onset of spring in the western United States. Bulletin of the American

663 Meteorological Society 82: 399-415.

664 Corn, P. C., and R. B. Bury. 1991. Terrestrial amphibian communities in the Oregon Coast

665 Range. Pages 305-317 in L. F. Ruggiero, K. B. Aubry, A. B. Carey, and M. H. Huff, Draft for Review Only – Not for Distribution 30 | Hayes et al. 666 technical coordinators. Wildlife and vegetation of unmanaged Douglas-fir forests. U.S.

667 Forest Service, General Technical Report PNW-GTR-285.

668 Crisafulli, C. M., L. S. Trippe, C. P. Hawkins, and J. A. MacMahon. 2005. Amphibian responses

669 to the 1980 eruption at Mount St. Helens. Pages 183-197 in V. H. Dale, F. J. Swanson,

670 and C. M. Crisafulli, editors. Ecological responses to the 1980 eruption of Mount St.

671 Helens. Springer Science+Business Media, Inc., New York, New York, USA.

672 Crother, B. I., J. Boundy, F. T. Burbrink, J. A. Campbell, K. De Quieroz, D. R. Frost, R.

673 Highton, J. B. Iverson, F. Kraus, R. W. McDiarmid, J. R. Mendelson III, P. A. Meylan, T.

674 W. Reeder, M. E. Seidel, S. G. Tilley, and D. B. Wake. 2008. Scientific and standard

675 English names of amphibians and reptiles of North America north of Mexico, with

676 confidence regarding confidence in our understanding. Sixth edition. Herpetological

677 Circular (37): 1-84.

678 Davis, T. M. 1998. Terrestrial salamander abundance in successional forests of coastal British

679 Columbia. Northwest Science 72:89-90.

680 Dumas, P. C. 1955. Eggs of the salamander Plethodon dunni in nature. Copeia 1955: 65.

681 Dumas, P. C. 1956. The ecological relations of sympatry in Plethodon dunni and Plethodon

682 vehiculum. Ecology 37:484-495.

683 Dupuis, L. A. 1997. Effects of logging on terrestrial amphibians of coastal British Columbia.

684 Herpetological Conservation 1: 185-190.

685 Dupuis, L. A., and F. Bunnell. 1999. Effects of stand age, size, and juxtaposition on abundance

686 of western red-backed salamanders (Plethodon vehiculum) in coastal British Columbia.

687 Northwest Science 73:27-33. Draft for Review Only – Not for Distribution 31 | Hayes et al. 688 Dupuis, L. A., J. N. M. Smith, and F. Bunnell. 1995. Relation of terrestrial-breeding amphibian

689 abundance to tree-stand age. Conservation Biology 9:645-653.

690 Dvornich, K. M., K. R. McAllister, and K. B. Aubry. 1997. Amphibians and reptiles of

691 Washington State: Location data and predicted distributions. Volume 2 in K. M. Cassidy,

692 C. E. Grue, M. R. Smith, and K. M. Dvornich, editors. Washington Cooperative Fish and

693 Wildlife Research Unit, University of Washington, Seattle.

694 Eaton-Mordas, A., E. P. Urling, M. P. Hayes, D. J. Dugger, and T. Quinn. 2004. Plethodon

695 dunni, Plethodon vehiculum (Dunn’s salamander, Western Red-backed Salamander).

696 Behavior. Herpetological Review 35:366-367.

697 Feder, M. E., and W. W. Burggren. 1992. Environmental physiology of the amphibians.

698 University of Chicago Press, Chicago, Illinois, USA.

699 Fitch, H. S. 1936. Amphibians and reptiles of the Rogue River Basin, Oregon. American

700 Midland Naturalist 17: 634-652.

701 Franklin, J. F., and C. T. Dyrness. 1988. Natural vegetation of Oregon and Washington. Oregon

702 State University Press, Corvallis, Oregon, USA.

703 Gilbert, F.F., and R. Allwine. 1991. Terrestrial amphibian communities in the Oregon Cascade

704 Range. Pages 319-324 in L. F. Ruggiero, K. B. Aubry, A. B. Carey, and M. H. Huff,

705 technical coordinators. Wildlife and vegetation of unmanaged Douglas-fir forests. U.S.

706 Forest Service, General Technical Report PNW-GTR-285.

707 Gnaedinger, L. M., and C. A. Reed. 1948. Contribution to the natural history of the plethodontid

708 salamander Ensatina eschscholtzii. Copeia 1948: 187-196.

709 Gomez, D. M., and R. G. Anthony. 1996. Amphibian and reptile abundance in riparian and

710 upslope areas of five forest types in western Oregon. Northwest Science 70: 109-119. Draft for Review Only – Not for Distribution 32 | Hayes et al. 711 Grialou, J. A., S. D. West, and R. N. Wilkins. 2000. The effects of forest clearcut harvesting and

712 thinning on terrestrial salamanders. Journal of Wildlife Management 64: 105-113.

713 Grove, S. J. 2002. Saproxylic insect ecology and the sustainable management of forests. Annual

714 Review of Ecology and Systematics 33:1-23.

715 Hagar, J. C., W. C. McComb, and C. C. Chambers. 1995. Chapter 9. Effects of forest practices

716 on wildlife. Cumulative effects of forest practices in Oregon: Literature and synthesis.

717 Oregon Department of Forestry, Salem, Oregon, USA.

718 Hanlin, H. G., J. J. Beatty, and S. W. Hanlin. 1978. A nest site of the western red-backed

719 salamander Plethodon vehiculum (Cooper). Journal of Herpetology 13: 214-216.

720 Heatwole, H. R. 1962. Environmental factors influencing local distribution and activity of the

721 salamander Plethodon cinereus. Ecology 43:460-472.

722 Herrington, R. E. 1988. Talus use by amphibians and reptiles in the Pacific Northwest. Pages

723 216-221 in R. C. Szaro, K. E. Severson, and D. R. Patton, technical coordinators.

724 Proceedings of the symposium on management of amphibians, reptiles, and small

725 mammals in North America. U.S. Forest Service General Technical Report RM-166.

726 Jackson, C. R., and C. A. Sturm. 2002. Woody debris and channel morphology in first- and

727 second-order forested channels in Washington’s Coast Ranges. Water Resources

728 Research 38:1177, doi:10.1029/2001WR001138.

729 Jayme, G. 1958. Properties of wood celluloses: II. Determination and significance of water

730 retention value. Tappi 41:180A.

731 Jones, L. L. C. 1989. Plethodon vandykei (Van Dyke’s salamander). Reproduction.

732 Herpetological Review 20: 48. Draft for Review Only – Not for Distribution 33 | Hayes et al. 733 Jones, L. L. C. 1999. Chapter VI. Survey protocol for the Van Dyke’s salamander (Plethodon

734 vandykei). Pp. 201-252 in D. H. Olson (editor), Standardized survey protocol for

735 amphibians under the Survey and Manage and Buffer Protection Provisions of the

736 Northwest Forest Plan. Version 3.0. U.S. Department of Agriculture, Forest Service,

737 Pacific Northwest Research Station, Olympia, Washington, USA.

738 Jones, L. L. C., W. P. Leonard, and D. H. Olson. 2005. Amphibians of the Pacific Northwest.

739 Seattle Audubon Society, Seattle, Washington, USA.

740 Mathieu, J., J.-P. Rossi, P. Mora, P. Lavelle, P. F. da S. Martins, C. Rouland, and M. Grimaldi.

741 2005. Recovery of soil macrofauna communities after forest clearance in eastern

742 Amazonia, Brazil. Conservation Biology 19:1598-1605.

743 Maxcy, K. A., and J. Richardson. 2000. Abundance and movements of terrestrial salamanders in

744 second-growth forests of southwestern British Columbia. Pages 295-302 in L. M.

745 Darling, editor. Proceedings of a conference of the biology and management of species

746 and habitats at risk, Kamloops, B.C., 15-19 February 1999, Volume One. British

747 Columbia Ministry of Environment, Lands, and Parks, Victoria, and University College

748 of the Cariboo, Kamloops, British Columbia, Canada.

749 McComb, W. C., C. L. Chambers, and M. Newton. 1993. Small mammal and amphibian

750 communities and habitat associations in red alder stands, central Oregon Coast Range.

751 Northwest Science 67: 183-189.

752 McGarigal, K., S. Cushman, and S. G. Stafford. 2000. Multivariate statistics for wildlife and

753 ecology research. Springer-Verlag New York, Inc., New York, USA. 283 pp.

754 McIntyre, A. P. 2004. Ecology of populations of Van Dyke’s salamander in the Cascade Range

755 of Washington State. Thesis, Oregon State University, Corvallis, Oregon, USA. Draft for Review Only – Not for Distribution 34 | Hayes et al. 756 McIntyre, A. P., C. M. Crisafulli, and R. A. Schmitz. 2005. Population and movements estimates

757 of Van Dyke’s salamanders (Plethodon vandykei) using mark-recapture techniques.

758 Northwestern Naturalist 86: 108.

759 McIntyre, A. P., R. A. Schmitz, and C. M. Crisafulli. 2006. Associations of the Van Dyke’s

760 salamander (Plethodon vandykei) with geomorphic conditions in headwall seeps of the

761 Cascade Range in Washington State. Journal of Herpetology 40: 309-322.

762 Nauman, R. S., D. H. Olson, L. L. Ellenburg, and B. P. Hansen. 1999. Plethodon dunni (Dunn’s

763 salamander). Reproduction. Herpetological Review 30: 89-90.

764 Norman, B. R. 1986. Ensatina eschscholtzii oregonensis (Oregon ensatina). Reproduction.

765 Herpetological Review 30: 89-90.

766 Norman, C. E., and B. R. Norman. 1980. Notes on the egg clusters and hatchlings of Ensatina

767 eschscholtzii oregonensis. Bulletin of the Chicago Herpetological Society 15: 99-100.

768 Ódor, P., J. Heilmann-Clausen, M. Christensen, E. Aude, K. W. van Dort, A. Piltaver, I. Siller,

769 M. T. Veerkamp, R. Walleyn, T. Standovár, A. F. M. van Hees, J. Kosec, N. Matocec, H.

770 Kraigher, and T. Grebenc. 2001. Diversity and composition of dead wood inhabiting

771 fungal and bryophytes communities in semi-natural beech forests in Europe. Nature-

772 based Management of Beech in Europe, Nat-Man Working Report 32.

773 Olson, D. H., R. S. Nauman, L. L. Ellenburg, B. P. Hansen, and S. S. Chan. 2006. Ensatina

774 eschscholtzii nests at a managed forest site in Oregon. Northwestern Naturalist 87: 203-

775 208.

776 Olson, D. H., P. D. Anderson, C. A. Frissell, H. H. Welsh, Jr., and D. F. Bradford. 2007.

777 Biodiversity management approaches for stream-riparian areas: Perspectives for Pacific

778 Northwest headwater forests, microclimates, and amphibians. Forest Ecology and Draft for Review Only – Not for Distribution 35 | Hayes et al. 779 Management 246:81-107.

780 Olson, D. H., and G. Weaver. 2007. Vertebrate assemblages associated with headwater

781 hydrology in western Oregon managed forests. Forest Science 53:343-355.

782 Ovaska, K. 1988. Spacing and movements of the salamander Plethodon vehiculum.

783 Herpetologica 44: 377-386.

784 Ovaska, K., and P. T. Gregory. 1989. Population structure, growth and reproduction in a

785 Vancouver Island population of the salamander Plethodon vehiculum. Herpetologica

786 45:133-143.

787 Robert, J. P. and T. A. Spies. 1998. Distribution of herbs and shrubs in relation to landform and

788 canopy cover in riparian forests of coastal Oregon. Canadian Journal of Botany 76:298–

789 315.

790 Raphael, M. G., P. A. Bisson, L. L. C. Jones, and A. D. Foster. 2002. Effects of streamside forest

791 management on the composition and abundance of stream and riparian fauna of the

792 Olympia Peninsula. Pages 27-40 in A. C. Johnson, R. W. Haynes, and R. A. Monserud,

793 editors. Proceedings from the Wood Compatibility Initiative Workshop, Number 3. U.S.

794 Forest Service, General Technical Report PNW-GTR-563.

795 Sollins, P. 1982. Input and decay of coarse woody debris in coniferous stands in western Oregon

796 and Washington. Canadian Journal of Forest Research 12: 18-28.

797 Stamm, A. J. 1935. Shrinking and swelling of wood. Industrial Engineering and Chemistry

798 27:401-406.

799 Stamm, A. J., and W. K. Loughborough. 1942. Variation in shrinking and swelling of wood.

800 Transactions of American Society of Mechanical Engineering 64:379-386. Draft for Review Only – Not for Distribution 36 | Hayes et al. 801 Vesely, D. G., and W. C. McComb. 2002. Salamander abundance and amphibian species

802 richness in riparian buffer strips in the Oregon Coast Range. Forest Science 48:291-297.

803 Walnick, R. 1997. Effects of forestry practices on amphibian populations in eastern North

804 America. Pages 191-205 in D. M. Green, editor. Amphibians in decline: Canadian studies

805 of a global problem. Herpetological Conservation I, Society for the Study of Amphibians

806 and Reptiles, St. Louis, Missouri.

807 Wentworth, C. K. 1922. A scale of grade and class terms for clastic sediments. Journal of

808 Geology 30:377-392.

809 Wilkins, R. N., and N. P. Peterson. 2000. Factors related to amphibian occurrence and abundance

810 in headwater streams draining second-growth Douglas-fir forests in southwestern

811 Washington. Forest Ecology and Management 139: 79-91.

812 Wilson, A. G., Jr., J. H. Larsen, and K. R. McAllister. 1995. Distribution of Van Dyke’s

813 salamander (Plethodon vandykei Van Denburgh). American Midland Naturalist 134: 388-

814 393.

815 Associate editor: Draft for Review Only – Not for Distribution

Figure Captions

Figure 1 – Map of the Willapa Hills portion of the Coast Range Physiographic Province,

Washington, USA showing the sites (black symbols) sampled in 2001 (circles) and 2003

(squares).

Figure 2 – Diagram of the streambank survey design relative to the stream and valley wall break.

The cross-section (A) shows one belt in profile; the aerial view (B) shows the relative

orientation of 4 adjacent belts. Each stream had 10 such belts.

Figure 3 – Varimax-rotated PCA analysis of physical and vegetation characteristics of 19 sites

sampled for terrestrial salamanders in the Willapa Hills, Washington, USA, 2001 and

2003. First principal component (Factor 1) scores increased with levels of small cobble,

bedrock, and boulders and decreased with increasing canopy cover; The second principal

component (Factor 2) increased with levels of sand and decreased with increasing moss.

Figure 4 – Variation in regional precipitation in the Willapa Hills region of southwest

Washington State, 2001 and 2003 for January-March (A; the 3 months prior to sampling

in this study) and April-June (B; the 3 months during sampling).

Figure 5 – Terrestrial salamander densities as a function of distance from the stream at Willapa

Hills region of southwest Washington State, 2001. Boxes delineate the 25th and 75th

percentiles with the median (if different) is a horizontal line within the box, the whiskers

are the tenth and ninetieth percentiles, and points are outliers beyond the 10th and 90th

percentiles. This summary includes only the 7 sites in which all 4 species were found in

2001. Draft for Review Only – Not for Distribution

Table 1. Amphibian species data for stream bank surveys at 19 sites across the Willapa Hills, southwest Washington, 2001, 2003. For each year, data are number of sites at which we recorded the species, and the means ( x ) and standard deviations (SD) of the number of belts per site where we found the species and the observed densities across sites (2001: n = 14; 2003: n = 5). We recorded 2 species of giant salamanders (Coastal giant salamander, Dicamptodon tenebrosus, and Cope’s giant salamander, D. copei), but we could not distinguish all individuals to species so we present summary data for both species collectively.

______

Occupancy

______

Sites No. Belts/Site Density (animals/m2)

______

2001 2003 2001 2003 2001 2003

______

Standard English Name Scientific Name n n x (SD) x (SD) x (SD) x (SD) ______

Ensatina Ensatina eschscholtzii 11 5 2.7 (2.2) 6.2 (1.6) 0.0035 (0.0030) 0.0090 (0.0048) Dunn's salamander Plethodon dunni 11 3 2.4 (1.6) 2.6 (3.0) 0.0026 (0.0021) 0.0059 (0.0088) Van Dyke's salamander Plethodon vandykei 10 1 1.7 (1.6) 0.2 (0.4) 0.0019 (0.0021) 0.0002 (0.0004) Western red-backed salamander Plethodon vehiculum 14 5 6.9 (2.0) 9.4 (1.3) 0.0152 (0.0085) 0.0382 (0.0230) Draft for Review Only – Not for Distribution 39 | Hayes et al. Coastal tailed frog Ascaphus truei 10 1 0.9 (0.7) 0.4 (0.9) 0.0009 (0.0007) 0.0003 (0.0008) Giant salamanders Dicamptodon spp. 8 4 1.4 (1.8) 1.2 (0.8) 0.0015 (0.0020) 0.0010 (0.0008) Columbia torrent salamander Rhyacotriton kezeri 11 3 2.1 (2.0) 1.4 (1.3) 0.0045 (0.0050) 0.0012 (0.0013) Northwestern salamander Ambystoma gracile 4 2 0.3 (0.5) 0.6 (0.9) 0.0002 (0.0004) 0.0005 (0.0007) Pacific chorus frog Pseudacris regilla 1 1 0.1 (0.3) 0.2 (0.4) 0.0001 (0.0002) 0.0002 (0.0004) Northern red-legged frog Rana aurora 2 1 0.1 (0.4) 0.2 (0.4) 0.0001 (0.0003) 0.0002 (0.0004) Rough-skinned newt Taricha granulosa 3 4 0.3 (0.6) 2.2 (2.3) 0.0003 (0.0005) 0.0020 (0.0023) ______

Draft for Review Only – Not for Distribution

Table 2. Variation in occupancy for the 4 terrestrial salamander species in banks and

uplands among 19 sites in the Willapa Hills, southwest Washington, 2001, 2003. Data are

means and standard deviations of occupied belts/site by species and year. We used

Friedman tests to compare species for each macrohabitat (rows); and Wilcoxon signed

rank tests to compare between those macrohabitats by species (columns). Chi-square

values comparing species had 3 degrees of freedom in 2001 but only 2 degrees of

freedom in 2003 because PLVA were too few to include the comparison. NC indicates

too few data for comparison.

______

Species Comparison

______

2 Year Macrohabitat ENES PLDU PLVA PLVE χ P ______

2001 Banks 0.6±1.1 2.1±1.9 1.5±1.4 4.1±2.1 22.8 <0.001

Uplands 2.1±1.7 0.3±0.7 0.1±0.3 5.1±1.5 33.7 <0.001

Comparison Z = −2.724 −2.680 −2.859 −2.232

P = 0.007 0.007 0.004 0.026

2003 Banks 1.8±0.6 2.4±2.8 0.2±0.5 7.6±2.8 12.7 0.005

Uplands 5.6±2.4 0.4±1.0 – 7.3±3.6 14.3 0.003

Comparison Z = −2.023 −1.604 NC 0.000

P = 0.043 0.109 NC >0.999

______Draft for Review Only – Not for Distribution 41 | Hayes et al. Table 3. Density (individuals/m2) for terrestrial salamanders in macrohabitat (banks and uplands)

from 19 sites in the Willapa Hills, southwest Washington, 2001, 2003. Data are means and

standard deviations. We used Friedman tests to compare species for each macrohabitat (rows);

and Wilcoxon signed rank tests to compare densities between macrohabitats (columns). Chi-

square values comparing species had 3 degrees of freedom in 2001 but only 2 degrees of

freedom in 2003 because PLVA were too few to include the comparison. NC indicates too few

data for comparison.

______

Species Comparison

______

2 Year Macrohabitat ENES PLDU PLVA PLVE χ P ______

2001 Banks 0.002±0.008 0.007±0.163 0.008±0.026 0.020±0.034 57.3 <0.001

Uplands 0.005±0.010 <0.001±0.002 0.001±0.001 0.012±0.017 131.0 <0.001

Comparison Z = −2.258 −4.684 −4.167 −1.576

P = 0.024 <0.001 <0.001 0.115

2003 Banks 0.004±0.011 0.012±0.029 0.001±0.006 0.044±0.044 77.1 <0.001

Uplands 0.012±0.014 0.001±0.004 – 0.037±0.031 85.4 <0.001

Comparison Z = −2.877 −3.181 NC -0.794

P = 0.004 0.002 NC 0.427

______

Draft for Review Only – Not for Distribution 42 | Hayes et al.

Table 4. Substrate temperature (° C) of microhabitats used by terrestrial salamanders (n) in the

Willapa Hills, southwest Washington, 2001 and 2003. Data come from 14 sites in 2001 and 5 sites in 2003 and are based on temperatures at the point where each animal was found.

______

Banks Uplands

______

Species Year n x SD n x SD

______

ENES 2001 10 11.2 2.9 41 10.7 2.4

2003 10 11.8 3.4 43 11.7 3.9

PLDU 2001 32 9.8 2.2 5 10.5 1.6

2003 28 13.1 2.6 6 11.2 1.8

PLVA 2001 25 9.8 1.6 1 12.5 –

2003 1 8.8 – 0 – –

PLVE 2001 105 10.6 2.2 115 10.7 2.3

2003 92 12.5 3.0 127 11.0 2.9

______Draft for Review Only – Not for Distribution

Table 5. Variation in cover and substrate categories used by the 4 terrestrial salamander species from the Willapa Hills, southwestern

Washington, 2001 and 2003. Data are weighted means and standard deviations of the proportion in each cover or substrate category from 14 sites in 2001 and 5 sites in 2003. Means were weighted based on the number of animals per site. Animal sample sizes (na =) are the number of animals on which weighted means are based. Difference in sample sizes between cover and substrate in a given row reflect the fact that some individuals were not found under cover.

______

Cover Substrates

______

Species Year na = Fines Inorganics Organics na = Fines Inorganics Organics

______

ENES 2001 Banks 10 0.20±0.04 – 0.80±0.04 10 0.60±0.25 – 0.40±0.25

2001 Uplands 42 0.10±0.11 0.02±0.01 0.88±0.11 44 0.52±0.11 0.07±0.04 0.41±0.13

2003 Banks 9 0.11±0.21 – 0.89±0.21 10 0.60±0.19 – 0.40±0.19

2003 Uplands 42 0.12±09 – 0.88±0.09 44 0.48±0.07 0.05±0.00 0.48±0.04

PLDU 2001 Banks 35 0.09±0.11 0.37±0.09 0.54±0.11 36 0.33±0.12 0.33±0.09 0.33±0.10 Draft for Review Only – Not for Distribution 44 | Hayes et al.

2001 Uplands 5 – – 1.00 5 0.20±0.36 0.40±0.09 0.40±0.09

2003 Banks 31 0.13±0.02 0.45±0.19 0.42±0.19 33 0.45±0.13 0.39±0.10 0.15±0.36

2003 Uplands 6 0.33±0.47 – 0.67±0.47 6 1.00 – –

PLVA 2001 Banks 30 – 0.10±0.13 0.90±0.13 30 0.20±0.04 0.13±0.13 0.67±0.11

2001 Uplands 1 – 1.00 – 1 – 1.00 –

2003 Banks 1 – – 1.00 1 – – 1.00

2003 Uplands 0 – – – 0 – – –

PLVE 2001 Banks 108 0.03±0.00 0.29±0.08 0.69±0.08 111 0.46±0.07 0.21±0.04 0.32±0.11

2001 Uplands 118 0.08±0.02 0.08±0.03 0.84±0.04 119 0.34±0.09 0.18±0.04 0.48±0.06

2003 Banks 103 0.14±0.00 0.19±0.02 0.67±0.02 105 0.45±0.02 0.21±0.03 0.34±0.08

2003 Uplands 126 0.14±0.01 0.05±0.01 0.81±0.02 130 0.32±0.01 0.14±0.03 0.54±0.05

______Draft for Review Only – Not for Distribution 45 | Hayes et al.

Table 6. Cover and substrates used by 4 terrestrial salamander species from the Willapa Hills, southwestern Washington, 2001 and

2003. Data are weighted means and standard deviations of the proportion in each cover or substrate category from a sample of 14 sites in 2001 and 5 in 2003. Means were weighted based on the number of animals per site. Animal sample sizes (na =) are the number of animals on which weighted means are based. Difference in sample sizes between cover and substrate in a given row reflect some individuals not being found under cover.

______

Cover Substrates

______

Species Year na = Large Wood Other Wood Non-Wood na = Large Wood Other Wood Non-Wood

______

ENES 2001 Banks 10 0.10±0.17 0.60±0.18 0.30±0.10 10 0.30±0.20 0.20±0.16 0.50±0.30

2001 Uplands 42 0.14±0.16 0.38±0.10 0.48±0.16 44 0.11±0.17 0.18±0.04 0.71±0.16

2003 Banks 9 0.00 0.33±0.17 0.67±0.17 10 0.00 0.00 1.00±0.00

2003 Uplands 42 0.05±0.00 0.46±0.02 0.49±0.04 44 0.07±0.00 0.25±0.03 0.68±0.05

PLDU 2001 Banks 35 0.17±0.03 0.26±0.07 0.57±0.12 36 0.14±0.02 0.11±0.03 0.75±0.04 Draft for Review Only – Not for Distribution 46 | Hayes et al.

2001 Uplands 5 0.20±0.09 0.40±0.34 0.40±0.27 5 0.20±0.09 0.00 0.80±0.09

2003 Banks 31 0.00 0.17±0.35 0.83±0.35 33 0.00 0.15±0.36 0.85±0.36

2003 Uplands 6 0.00 0.33±0.03 0.67±0.03 6 0.00 0.00 1.00

PLVA 2001 Banks 30 0.27±0.05 0.30±0.16 0.43±0.14 30 0.37±0.14 0.20±0.05 0.43±0.11

2001 Uplands 1 0.00 0.00 1.00 1 0.00 0.00 1.00

2003 Banks 1 1.00 0.00 0.00 1 1.00 0.00 0.00

2003 Uplands 0 0.00 0.00 0.00 0 0.00 0.00 0.00

PLVE 2001 Banks 108 0.06±0.03 0.36±0.11 0.58±0.12 111 0.11±0.02 0.13±0.06 0.77±0.08

2001 Uplands 118 0.10±0.02 0.36±0.06 0.54±0.06 119 0.13±0.03 0.19±0.04 0.68±0.05

2003 Banks 103 0.15±0.01 0.25±0.04 0.60±0.02 105 0.06±0.01 0.22±0.02 0.72±0.04

2003 Uplands 126 0.10±0.02 0.48±0.02 0.42±0.03 130 0.10±0.01 0.32±0.02 0.58±0.04

______Draft for Review Only – Not for Distribution

Table 7. Mean differences in moisture and temperature between used versus available habitat for terrestrial salamanders from the Willapa Hills, southwest Washington, 2001 and 2003. Used habitat was defined as the temperature at the location of the animal and available habitat was the temperature from the adjacent macrohabitat (see text for details). Data are means ( x ) and standard deviations (SD) of the differences between used and available moisture or temperature values. Where sample sizes differ, the first refers to moisture and the second for temperature.

Negative means indicate that used values averaged less than available values. Probabilities (P) are for Wilcoxon signed rank tests applied to the H0: Used – available habitat = 0.

______

Moisture Temperature ______

Species Year Macrohabitat n = x SD P x SD P ______

ENES 2001 Bank 10 -0.30 0.48 0.083 0.18 1.98 0.776

PLDU 2001 Bank 36,34 0.28 0.57 0.008 -1.09 1.56 <0.001

PLVA 2001 Bank 30 0.27 0.64 0.034 -0.24 1.62 0.421

PLVE 2001 Bank 111,110 -0.03 0.39 0.467 -0.53 1.69 0.002

ENES 2003 Bank 10 0.00 0.24 >0.999 0.11 1.82 0.860

PLDU 2003 Bank 34,33 -0.02 0.31 0.706 0.09 2.66 0.841

PLVE 2003 Bank 105 0.04 0.36 0.356 -0.24 1.95 0.220

ENES 2001 Upland 41,38 0.02 0.16 >0.999 -1.00 1.59 <0.001

PLVE 2001 Upland 119 0.13 0.38 <0.001 -0.85 1.43 <0.001

ENES 2003 Upland 45 0.02 0.38 0.558 -0.45 1.19 0.015

PLVE 2003 Upland 130 0.04 0.23 0.059 -0.03 1.31 0.831 ______Draft for Review Only – Not for Distribution 48 | Hayes et al. Table 8. Spearman rank correlations (ρ) between the proportions of animals utilizing wood (by

type) as cover versus the amount of wood at that site for terrestrial salamanders on streambanks

in the Willapa Hills, southwest Washington, 2001. Functional wood is LWD in Sollins (1982)

decay classes 2, 3, or 4; LWD50 is LWD >50 cm in diameter in decay classes 2, 3, or 4.

______

PLDU PLVA PLVE

______

Wood Measure ρ P ρ P ρ P

______

LWD Density (pieces/m2) -0.599 0.072 0.756 0.032 0.097 0.725

Functional LWD Volume (m3/m2) -0.488 0.143 0.513 0.147 -0.128 0.643

Functional LWD Density (pieces/m2) -0.272 0.415 0.319 0.366 0.233 0.402

3 2 LWD50 Volume (m /m ) -0.636 0.056 0.294 0.406 -0.230 0.406

2 LWD50 Density (pieces/m ) -0.679 0.042 0.160 0.652 0.060 0.829

______Draft for Review Only – Not for Distribution 49 | Hayes et al. Appendix 1. Categories used in habitat evaluation. Analysis groupings were fines (1), inorganics

(2), and organics (3); and large wood (4), other wood (5), and non-wood (6); see text for details.

______

Categories Definitions Grouping ______

Fines ≤ 2 mm particle size diam 1 6

Inorganic Substrates

Gravel > 2-16 mm particle size diam 2 6

Small Cobble 17-64 mm particle size diam 2 6

Large Cobble 65-256 mm particle size diam 2 6

Boulder > 256 mm particle size diam 2 6

Bedrock particle size too large or unrecognizable 2 6

Organic Substrates

Large Woody Debris (LWD):

Small 10-25 cm cross-sectional diam 3 5

Medium 26-50 cm cross-sectional diam 3 5

Large > 50 cm cross-sectional diam 3 4

Other Wood logs < 10 cm diam; any other wood 3 5

Leaf litter leaf and conifer needle debris 3 6

Live vegetation:

Moss moss or moss carpeting 3 6

Non-moss vegetation non-woody, non-moss vegetation 3 6

______

Draft for Review Only – Not for Distribution

Draft for Review Only – Not for Distribution

Valley Wall Break

L Stream R U A Channel Left Bank Right Bank Uplands (Variable length) (Variable length) (35 m) Draft for Review Only – Not for Distribution

Valley Wall Break

L R U

U L R

Stream Channel L R U

U L R B Draft for Review Only – Not for Distribution

3.0 2001 2.0 2003

1.0

0.0 Factor 2 Factor

-1.0

-2.0

-3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 Factor 1

1400

1200 A January-March 2001 1000 2003

800

600 Precipitation (mm) Precipitation

400

200

0 Centralia Naselle Raymond Toledo

1400

1200 B April-June 2001 1000 2003

800

600 Precipitation (mm) Precipitation

400

200

0 Centralia Naselle Raymond Toledo Draft for Review Only – Not for Distribution

0.08 ENES 0.06

0.04

0.02

0.00 0-6 7-12 13-18 19-24 25-30 31-36 0.08

0.06 PLDU

0.04

) 2 0.02

0.00 0-6 7-12 13-18 19-24 25-30 31-36 0.08 PLVA 0.06 Density (individuals/m

0.04

0.02

0.00 0-6 7-12 13-18 19-24 25-30 31-36 0.08

0.06 PLVE

0.04

0.02

0.00 0-6 7-12 13-18 19-24 25-30 31-36 Distance from Stream (m)